Reference
Roadmap
An honest map of what works today versus what is designed but not yet shipped. Everything documented elsewhere reflects verified, live behavior.
Live today#
- REST API:
health,context,scopes(list + create/get/delete),search,answer,units,units/:id,units:delete,catalog/stats,ingest,compaction,audit,audit/verify,benchmarks,eval,admin/seed. - Registries:
grantsandprincipalssupport create/list/get/delete (admin);namespacesare self-service (list + create your own). - Multi-namespace tenancy: own several namespaces, switch with
X-Aegis-Namespace, auto-provisioned own namespace + starter scope on sign-up, requested-scope binding for reads/writes. - Billing (Stripe): Free/Pro/Enterprise tiers × monthly/yearly/lifetime — live plans, checkout, subscription, portal, reconcile, and a signature-verified webhook that fulfills tiers.
- Tier gating: answer/rerank/multimodal-planes/audit feature gates, document ceiling, and namespace quota (graceful clamp on search; hard 402 elsewhere).
- Multimodal hybrid search: image/audio/video/doc across the text, visual, audio, and doc-visual planes; dense (vector index) + lexical (catalog FTS5), RRF fusion, cross-encoder rerank, sub-unit Locators, cost/latency trace.
- Grounded cited answers via
POST /v1/answer: a Workers AI LLM (primaryllama-3.1-8b-instruct-fp8→ CFllama-3.3-70bfallback → OpenAI fallback → extractive), answering strictly from cited evidence with a groundedness score. - Security: RBAC/ABAC, scope clamping, grant intersection, per-hit ACL filtering, and a hash-chained audit log with independent verification.
- Developer surfaces: TypeScript + Python SDKs, the CLI, and the agent-native MCP server.
- Web console: sign-in (password / email code / magic link / Google), workspace overview, search playground, unit inspector, namespace & scope switcher, tier-aware UI, upgrade/billing page, branded 404.
- Multimodal ingestion via
POST /v1/ingest(modality + plane + Locator; video → audio → ASR pipeline).
Coming soon#
Not available yet
The following are part of the design but are not implemented today — documented honestly so you don't build against them. Until they ship, use the REST API directly.
| Area | Status | Notes |
|---|---|---|
| Per-user API keys | In progress | Console-minted, revocable per-user tokens; use seeded/server-issued tokens today. |
| Agent & Query-the-DB console tabs | Planned | Conversational agent surface and the saved query builder are stubs (nav marked coming-soon). |
| Saved queries & dashboards | Planned | “Save query” / “add to dashboard” are inert placeholders today. |
| Crawlers, Schedulers & Compute | Planned | Automated ingestion crawlers, scheduled jobs, and a compute surface are designed, not built. |
| Observability & Integrations | Planned | In-console metrics dashboards and third-party integrations/settings are stubs. |
| Streaming answers & MLLM-judge verify | Planned | Answers are buffered JSON; faithfulness is a groundedness proxy, not full NLI entailment yet. |
| PATCH updates & raw source fetch | Planned | In-place metadata updates and GET /v1/sources/:id return 404 today. |
| SSO / Access (JWT) auth | Reserved | The X-Access-Jwt-Assertion path is not yet verified; use Bearer tokens or console sessions. |
API stability#
The contract is versioned under /v1. Response shapes documented here are stable for building against; treat similarity scores as informational (they shift as models change) and assert on invariants like ordering, membership, and ACL counts instead.